Physical Chemistry Chemical Physics
Core Structure Dependence of Cycloreversion Dynamics in Diarylethene Analogs
Journal: Physical Chemistry Chemical Physics
Manuscript ID CP-ART-10-2019-005797.R1
Article Type: Paper
Date Submitted by the 09-Jan-2020 Author:
Complete List of Authors: Honick, Chana; Johns Hopkins University, Chemistry Peters, Garvin; Johns Hopkins University, Chemistry Young, Jamie; Johns Hopkins University, Chemistry Tovar, John; Johns Hopkins University, Chemistry Bragg, Arthur; Johns Hopkins University, Chemistry
Page 1 of 43 Physical Chemistry Chemical Physics
Core Structure Dependence of Cycloreversion Dynamics in Diarylethene Analogs
Chana R. Honick, Garvin M. Peters, Jamie D. Young, John D. Tovar, and Arthur E. Bragg* Department of Chemistry, Johns Hopkins University 3400 N. Charles St., Baltimore, MD 21218 *[email protected]
Abstract: Diarylperfluorocyclopentenes are a well-characterized class of molecular
photoswitches that undergoes reversible photocyclization. The efficiency of cycloreversion
(<~30%), in particular, is known to be limited by a competition with excited-state deactivation by
internal conversion that is strongly impacted by the electron-withdrawing/donating character of
pendant aryl groups. Here we present a first study to determine how varied structural motifs for
the core bridge group impact excited-state dynamics that control cycloreversion quantum yields.
Specifically, we compare photophysical behaviors of 3,3'-(perfluorocyclopent-1-ene-1,2-
diyl)bis(2-methylbenzo[b]thiophene) with diarylethene derivatives possessing the same
benzo[b]thiophene pendant group but with a rigid 1-methyl-1H-pyrrole-2,5-dione and a
1 Physical Chemistry Chemical Physics Page 2 of 43
rigid/aromatic thieno[3,4-b]thiophenbridge (TT) core bridge group. We find that the flexible
perfluorocyclopentene core undergoes cycloreversion 3-4x slower than the rigid core
photoswitches (9 vs. 2-3 ps in acetonitrile, 25 vs. 5-6 ps in cyclohexane) despite comparable
cycloreversion quantum yields. To distinguish effects induced by bridge vs pendant groups, we
also studied a series of photoswitches with the same thieno[3,4-b]thiophene bridging group, but
with varied pendant groups including 2,5-dimethylthiophene and 2-(3,5-
bis(trifluoromethyl)phenyl)-5-methylthiophene. Analysis of temperature-dependent excited-state
lifetimes and cycloreversion quantum yields reveals that both the rates of nonreactive internal
conversion and reactive cycloreversion increase with greater structural rigidity of the core. This
difference is attributed to smaller energy barriers on the excited-state potential energy surface for
both reactive and non-reactive deactivation from the 21A electronic state relative to the flexible
perfluorocyclopentene switch, implying that a rigid core results in a net shallower excited-state
potential energy surface.
Introduction: Molecular photoswitching is possible when a molecule undergoes reversible
isomerization between two or more structures following isomer-selective absorption of light.
Photoswitches can be used to manipulate both optical and electronic material properties and have
been investigated widely for optoelectronic applications including optical memory storage and
optical gating of molecular electronics.1-7 For precision performance, it is imperative that photo- switches have high thermal stability, rapid response times, high switching efficiency (i.e. quantum yield), and fatigue resistance over many switching cycles.1, 2, 8-14 Activation with visible to near
infrared wavelength light is particularly advantageous for utilizing widely available and
inexpensive light sources (e.g. LEDs), as well as for improving penetration depths through
2 Page 3 of 43 Physical Chemistry Chemical Physics
materials and biological tissue.1 Diarylethenes (DAE) are a class of photoswitch that possess many
of these desirable characteristics. 1, 2, 10 For example, bis(benzo[b]thienyl)perfluorocyclopentene
(1o), shown in Scheme 1, undergoes electrocyclization following excitation of its lowest-energy
absorption while cycloreversion follows after selective visible excitation of the lowest absorption
of its cyclized isomer (1c). Notably, specific absorption characteristics of the open and closed
isomers depend on the structure (e.g. extent of conjugation across the cyclized isomer). Some
DAE derivatives have been shown to undergo rapid reversible cyclization several thousand times,
and in the absence of light, many can remain in one form or another indefinitely.15, 16
There has been extensive synthetic exploration of this class of compounds in recent years
with demonstrated improvements in isomerization quantum yields and fatigue resistance via
structural modifications to the pendant groups that form a bond during the cyclization.9, 10, 17-21
This work has been accompanied by a number of computational 22-26 and ultrafast spectroscopic
investigations27-35 to address how these modifications alter the shapes of and dynamics on excited-
state potential energy landscapes that underlie their switching efficiencies. For DAEs like 1o, the
origins of inefficiencies depend on switching direction. Photocyclization efficiency is generally
limited in solution by thermal population of both reactive parallel and non-reactive antiparallel
conformers2, 26, whereas switching efficiency can be significantly higher in crystals that favor the
reactive conformer as a result of molecular packing.13, 31, 36-38 Cyclization is also inefficient for
structures with large initial separations between the reactive carbon sites of the pendant rings.17
Thus, DAEs in a crystal with a C-C separation <4 Å exhibit cyclization quantum yields
approaching unity.
On the other hand, cycloreversion yields are comparatively lower for normal-type DAE
switches, usually much less than 0.5.22 The prevailing rationale for this behavior is that a
3 Physical Chemistry Chemical Physics Page 4 of 43
significant energetic barrier exists on the 2 1A excited-state surface that separates nonreactive and reactive conical intersections (CIs) for deactivation (i.e. “internal conversion” vs.
“cycloreversion”); this barrier is nominally associated with avoided crossings with higher-lying states of the same (A) symmetry.22, 25 Temperature-dependent interrogation of cycloreversion quantum yields and ultrafast excited-state deactivation dynamics in 1c, in particular, provided empirical support for this excited-state barrier,32 while computations reveal that the magnitude of this barrier, which dictates accessibility of the reactive CI for ring opening, is strongly dependent on the properties of pendant groups.22 Under these circumstances, increasing the efficiency of ring opening requires tuning or manipulating ultrafast dynamics on excited-state potential energy surfaces to favor transit to a CI for cycloreversion. Recent strategies towards the improvement of switching performance involve multi-photon induced and temperature-dependent photoswitching,29, 30, 32, 39-42 both of which require additional external stimuli. A more general strategy for improving switch performance is to differentiate the structure-dynamics relationships associated with these two processes to control their relative rates in favor of cycloreversion.
Scheme 1. Photoconversion between the cyclized (1c) and open (1o) isomers of 3,3'- (perfluorocyclopent-1-ene-1,2-diyl)bis(2-methylbenzo[b]thiophene), and structures of the closed isomers (c) of photoswitches investigated in this work: 3,3'-(thieno[3,4-b]thiophene-2,3- diyl)bis(2-methylbenzo[b]thiophene) (2o/2c), 1-methyl-3,4-bis(2-methylbenzo[b]thiophen-3-yl)- 1H-pyrrole-2,5-dione (3o/3c), 2,3-bis(5-(3,5-bis(trifluoromethyl)phenyl)-2-methylthiophen-3- yl)thieno[3,4-b]thiophene (4o/4c), and 2,3-bis(2,5-dimethylthiophen-3-yl)thieno[3,4-b]thiophene (5o/5c).
4 Page 5 of 43 Physical Chemistry Chemical Physics
Notably, prior efforts focused on the properties of switches that share a common
core/bridge motif, perfluorocyclopentene (e.g., 1o in Scheme 1). Indeed, perfluorocyclopentene
has been the most commonly studied DAE bridging group to date due to its stability and non-
reactivity. Ultrafast dynamics of cyclization and cycloreversion have been studied almost
exclusively with switches built from this bridging group, including photoswitches with various
pendent groups: 1,2-bis(2,4-dimethyl- 5-phenyl-3-thienyl),29, 30 1,2- bis(2 - methyl-5-(2 – (4 -
benzoyl-phenyl-vinyl))-thien- 3-yl),43 1,2-bis(5-formyl-2- methyl-thien-3-yl),33, 44 bis(2-methyl-5-
phenylthiophen-3-yl),39-41, 45 and 1,2-bis(2-methylbenzo[b]-thiophen-3-yl)32, 46, 47. All of these
5 Physical Chemistry Chemical Physics Page 6 of 43
switches have qualitatively similar cycloreversion dynamics in hexane or cyclohexane with initial spectral relaxation occurring on timescales ranging 1-4 ps that is attributed to fast structural evolution in the excited state, and with subsequent excited-state lifetimes ranging from 9 to 25 ps.
These relatively long excited-state lifetimes may enable deactivation pathways competitive with cycloreversion.
Although perfluorocyclopentene-core switches have demonstrated considerable success, other switch cores could exhibit useful function. For example, aromatic bridging groups have been considered synthetically for purposes of linking switchable chromophores to larger conjugated structures to enable photo-switchable electronic properties. We recently integrated a thienothiophene core (e.g. as in 2) into DAEs to facilitate synthesis of core-linked oligomers with electronic properties controllable through the cyclization of peripheral groups.48 It could be hypothesized that photoinduced cycloreversion may be favored with the introduction of core aromaticity due to an anticipated increase in the stability of the open isomer relative to a structure with a less globally aromatic core: e.g., in the case of 2, one would expect a difference in stability between the 6π aromatic system in the closed form and the 10π aromatic system in the open form.
Irie and coworkers have also shown that introducing an aromatic thiophene bridging group alters the pi-conjugation of both the open and closed-form isomers.4 Steady-state properties of these systems were altered as a result of extending conjugation through the bridging group, but the impact of an aromatic rigid core group on switching dynamics and efficiency has not been examined explicitly. A related study on excited-state topologies of dihydroazulene-photoswitches indicated that changes in aromaticity along the potential energy surface could significantly alter switching characteristics, particularly hampering switching when the excited transition-state involves a loss in aromaticity.49
6 Page 7 of 43 Physical Chemistry Chemical Physics
Notably, ultrafast switching dynamics may also be strongly affected by the inherent rigidity
of the core bridging group: this rigidity could significantly impact the potential energy landscape
and barriers for structural relaxation to critical excited-state geometries (i.e. conical intersections)
for nonadiabatic cycloreversion, thereby reducing the rate and efficiency of cycloreversion relative
to other deactivation processes. Alternatively, a rigid core could be expected to constrain or funnel
excited-state dynamics to critical structures, resulting in faster switching dynamics.50 This second
possibility is notable because the standard perfluorocyclopentene core structure is comparatively
flexible, which could facilitate structural evolution along excited-state potential-energy surfaces
to local excited-state minima resulting in somewhat longer excited-state lifetimes.
In order to address the role of core bridge structure on photoswitch properties, we present
an investigation of the ultrafast cycloreversion dynamics and quantum yields for a series of
photoswitches 1c-3c (Scheme 1) that contain various core/bridge motifs including perfluorinated
cyclopentene (1c), a rigid, aromatic thieno[3,4-b]thiophene (2c), and a rigid non-aromatic 1-
methyl-1H-pyrrole-2,5-dione (3c), but all with common benzo[b]thiophene pendant groups. We
also present the analysis of a series of photoswitches with a common thienothiophene core motif
but with various pendant groups (2c, 4c and 5c). By comparing photophysics of these five
compounds, we extend current understanding of excited-state dynamics in DAE photoswitches
and particularly the impact of core structure on the shape of the excited-state potential energy
surface that control these dynamics and cycloreversion quantum yields.
2. Experimental methods.
2.1 Sample Preparation and Handling. 3,3'-(perfluorocyclopent-1-ene-1,2-diyl)bis(2-
methylbenzo[b]thiophene) (1o) (> 97.0 %) was purchased from TCI America and used as
7 Physical Chemistry Chemical Physics Page 8 of 43
received. 3,3'-(thieno[3,4-b]thiophene-2,3-diyl)bis(2-methylbenzo[b]thiophene) (2o), 1-methyl-
3,4-bis(2-methylbenzo[b]thiophen-3-yl)-1H-pyrrole-2,5-dione (3o),51 2,3-bis(5-(3,5- bis(trifluoromethyl)phenyl)-2-methylthiophen-3-yl)thieno[3,4-b]thiophene (4o),48 and 2,3-
bis(2,5-dimethylthiophen-3-yl)thieno[3,4-b]thiophene (5o),48 were synthesized and purified as
described elsewhere; the synthetic details for 2 are described in the Supporting Information.
Sample solutions were prepared by dissolving 1o, 2o, 3o, and 4o both in acetonitrile and in
cyclohexane at concentrations ranging from ~0.6-20x10-3 M. A solution of 5o was only prepared
in acetonitrile at similar concentrations. Each sample was subsequently converted to a
photostationary state of open- and cyclized isomers through prolonged irradiation at 320 nm with
an ultraviolet hand lamp resulting in concentrations of the cyclized form at steady state ranging
~1-4 x 10-4 M. We include a spectrum of the hand lamp in the Supporting Information Figure S28.
Photostationary solutions were stored in the dark following preparation to inhibit photoinduced
cycloreversion.
Steady-state UV-Vis spectra of photoconverted samples were collected both before and
after time-resolved measurements using a diode-array spectrometer with fiber-optically coupled
tungsten and deuterium lamps (Stellarnet). To prevent a significant drop in concentration of the
cyclized isomer throughout the course of time-resolved measurements, each solution was
continuously pumped through a flow cell (variable pathlength from Harrick Scientific Products or
a 1mm quartz flowcell from Spectrocell) from a solution reservoir contained in a quartz round-
bottom flask. The quartz flask was irradiated with a UV hand lamp throughout the course of each
measurement to maintain a photostationary state. The optical path-length of the flow cell was either
0.5 mm, 0.95 mm, or 1 mm depending on the optical density of the closed isomer at excitation
wavelengths used for ultrafast measurements. Optical densities in the flow cell ranged from ~0.03
8 Page 9 of 43 Physical Chemistry Chemical Physics
to 0.3. A cryostat with a 1 cm cuvette was used for temperature-dependent transient absorption
and quantum yield measurements. Temperature-dependent ultrafast measurements were
conducted over a short acquisition timescale (~1 min.) and samples were stirred rapidly in order
to minimize sample loss due to photoconversion. A reference signal collected periodically at a
fixed pump-probe time delay was also collected in order to normalize transient signals with respect
to conversion as a result of irradiation of sample cuvettes with ultrafast pulses.
2.2 Transient Absorption Spectroscopy. A regeneratively amplified Ti:sapphire laser system
(Coherent Legend Elite, 1 kHz rep. rate, 35 fs pulse duration, 4.0 mJ/pulse) was used to generate
laser pulses for ultrafast spectroscopic measurements. Pump pulses centered at 500 nm, 530 nm,
and 600 nm were generated with an optical parametric amplifier (OPA, Coherent Operasolo) via
signal+800 nm sum-frequency generation with fluences ranging from ~28 to ~250 J/cm2; no
power dependence of dynamics were observed in this range. These wavelengths are selectively
resonant with the lowest energy 500 nm absorption features of 1c, 2c, and 3c, the 600 nm
absorption feature of 4c, and the 530 nm absorption feature of 5c (vide infra), ensuring the absence
of contaminating spectral signals in transient spectra that would be associated with photoinduced
cyclization of the open isomers. A broadband probe continuum (~380-700 nm) was generated by
focusing 800 nm ultrafast pulses through a calcium fluoride (CaF2) plate. A temporal window of -
10-1200 ps was achieved by retro-reflecting the pump beam off of a hollow corner cube mirror
mounted onto a motorized translation stage (Newport). To remove the effects of polarization
anisotropy, the probe beam was directed through a wire-grid polarizer set to magic angle relative
to the pump beam polarization. After passing through the sample, the probe beam was filtered to
remove features below 370 nm and above 800 nm and was dispersed and focused onto a
9 Physical Chemistry Chemical Physics Page 10 of 43
multichannel array. Probe spectra were collected and transient absorption subsequently calculated
shot-to-shot with a homebuilt LabVIEW program described elsewhere.52, 53 Our time resolution
(pump-probe cross-correlation) was approximately 120 fs for these measurements. However, large-amplitude coherent artifacts contributed significantly to signals measured within a few hundred femtoseconds of time zero. Time-resolved spectra collected at time delays from 0.5 ps and beyond were used in global fitting analyses, as we could not differentiate faster spectral dynamics from coherent signals at shorter delays.
2.3 Molar Absorptivity and Quantum Yield Determinations.
The molar absorptivities of 1o-5o were determined from a Beer’s Law analysis of steady- state UV-Vis absorption measurements with solutions at various photoswitch concentrations in a
1-mm cuvette. The molar absorptivities for 1c-5c in relation to 1o-5o were determined using chromatographic separations of photostationary mixtures of the isomers in combination with UV-
Vis absorption spectroscopy. Chromatographic separations used reverse-phase HPLC
(methanol/water eluent with a 00D-4725-AN Kinetex column from Phenomenex, particle size:
2.6 m, pore size: 100 Å, length: 100 mm for 1o/c, 2o/c, 3o/c, 4o/c, and 200 mm for 5o/c). The fraction of closed isomer within the photostationary mixture was determined based on the fractional drop in the open-isomer concentration measured before and after irradiation to the photostationary state. The relative absorptivities of the two isomers were determined at a fixed wavelength in the UV by normalizing the absorption spectra measured for each isomer after HPLC separation by the fraction of each isomer recovered; the absolute extinction coefficient for the closed isomer could then be determined relative to that of the open isomer. Details pertaining to
10 Page 11 of 43 Physical Chemistry Chemical Physics
the determination of molar absorptivities (including a list of determined values) are provided in
Section 2 of the Supporting Information.
The cycloreversion quantum yields of each system were determined using photochemical
actinometry.54, 55 We used the cycloreversion of 1,2-bis(2,4-dimethyl-5-phenyl-3-thienyl)-
3,3,4,4,5,5-hexafluoro-1-cyclopentene (98.0% TCI) as an actinometric standard because the
quantum yields had been determined previously at several excitation wavelengths in the range of
480-620 nm.14 This standard was used specifically to determine the photon flux of a 496 nm (cyan,
FWHM: 485-507nm) light emitting diode (LED) (Led World via Amazon, ASIN: B00MO9O7Q8)
used to drive isomerization of 1c, 2c, 3c, and 5c as well as a 596 nm (yellow FWHM:587-602nm)
LED (Sparkfun: Model No.: YSL-R1031Y2C-D14) used to drive isomerization of 4c. These LEDs
were chosen to match the maxima of the ground state absorption of the closed structures and also
because they do not overlap with absorption features of open isomers. Temperature-dependent
quantum yields for cycloreversion of 1c-3c were determined using a cryostat (Unisoku) by the
same general method. For these measurements, the quantum yield of each switch at 298 K was
used as a standard in order to determine the photon flux through the cryostat. The absorption
spectra of closed switches (without LED irradiation) were also measured at various temperatures
to account for any changes in molar absorptivity with temperature when determining quantum
yields. Full experimental details for acquiring and analyzing data for quantum-yield
determinations are described further in the Supporting Information.
3. Results.
3.1 Steady-state spectroscopic characterization.
11 Physical Chemistry Chemical Physics Page 12 of 43
The molar absorptivities (M-1cm-1) of 1-5 as both open and closed isomers are plotted in
Figure 1 (a-e); steps for determining these values are outlined in the Supporting Information (see, e.g. Figures S1-S6). The absorption spectra of 1o and 1c are similar to those published previously.32, 46, 47 The absorption onset for 2o, 4o and 5o are similar to that of 1o, whereas the lowest-energy absorption of 3o is shifted noticeably to a longer wavelength. The lowest-energy absorption features of 1c, 2c, 3c, and 5c fall near 500 nm, with differences in the positions of higher energy features. In contrast, the lowest-energy absorption of 4c falls to longer wavelength
(~600 nm). These spectral changes from open to closed structures are consistent with breaking aromaticity of the open structure via cyclization to create delocalized polyene network. These spectra also illustrate that quantum yield and ultrafast transient measurements conducted with light sources at 500 nm and longer wavelengths should be selective for excitation of the closed isomer.
12 Page 13 of 43 Physical Chemistry Chemical Physics
Figure 1. Molar absorptivities for the open and closed isomers of 1-5 in acetonitrile.
13 Physical Chemistry Chemical Physics Page 14 of 43
3.2 Ultrafast characterization of excited-state dynamics.
3.2.1 Core-dependence of excited-state dynamics (1c-3c).
Figures 2a, 2b, and 2c display transient absorption spectra collected following 500 nm
excitation of 1c-3c in acetonitrile as probed across the visible. For 1c, transient spectra closely resemble those reported by Shim et al., but with some differences in spectral evolution arising from differences in solvent environment that will be discussed further below. Here we observe a narrow absorption feature centered about 695 nm as well as a few weaker features at 634, 555, and
435 nm that overlap with the region of the lowest-energy steady-state absorption feature for 1c.
All absorption features decay within 100 ps; notably, a bleach of the ground-state absorption spectrum (“ground-state bleach,” or GSB) with peak near 530 nm appears at latest delays examined, consistent with the partial depletion of 1c through cycloreversion to 1o.
The transient spectra for 2c and 3c are somewhat different: At 0.5 ps, 2c exhibits a slight negative intensity at 515 nm with photoinduced absorption both at shorter and longer wavelengths.
As the lowest-energy feature in the ground-state absorption spectrum peaks roughly in the middle of this region, we interpret spectra at earliest delays to reflect a combination of both GSB and photoinduced absorption. Similar to 2c, 3c exhibits a broad absorptive feature extending across the visible spectrum, with the GSB overlapping the absorption at 502 nm that blueshifts slightly to 498 nm on a sub-picosecond timescale. Like 1c, 3c also exhibits an intense photoinduced absorption above 700 nm. 3c also exhibits a signature of a stimulated emission (SE) band at 665 nm that overlaps its excited-state absorption features. By the latest delays probed the photoinduced absorption (and emission) features have decayed and GSB appears near 500 nm, likewise indicating conversion of 2c and 3c to 2o and 3o, respectively. Contrary to 1c, 2c and 3c both
14 Page 15 of 43 Physical Chemistry Chemical Physics
exhibit a minor bleach recovery that is associated with the decay of an absorption just to the red
of the bleach; for reasons described below, we attribute this to relaxation of a vibrationally hot
ground electronic state.
Figure 2. Transient absorption spectra collected for 1c (a), 2c (b), and 3c (c) in acetonitrile at selected time delays ranging from 0.5ps to 100ps as well as the corresponding species-associated spectra (SAS) as determined through global fitting for 1c (d), 2c (e), and 3c (f). Data was normalized to 1 at maximum absorption at 1ps.
15 Physical Chemistry Chemical Physics Page 16 of 43
Figure 2d, 2e, and 2f plot the species associated spectra (SAS) obtained through principal component global analysis of TA data for 1c-3c subject to series, state-to-state kinetic models for all transient data collected after a delay of 0.5 ps. Figures 3 plots the time-dependent global fits for transients of different probe wavelengths across the visible (450 nm; 550 nm; 630 nm; and 700 nm) for each switch; spectral reconstructions from global fits and residuals are presented in Figures
S7-S9. Best-fit lifetimes obtained for each switch are summarized in Table 1. Transient data collected for 1c was fit with a three-state series model, yielding interconversion lifetimes of 0.5 ps and 8.9 ps. The shorter lifetime is associated with a rapid spectral shifting/relaxation of the initial excited state (흉풓풆풍풂풙 ) whereas the slower timescale is associated with the overall excited-state lifetime (흉풅풆풄풂풚 ퟏ ).
Figure 3. Time-dependent global fits for 1c (column a), 2c (column b), and 3c (column c) in acetonitrile. Data after 0.5 ps is plotted on a linear scale <1ps and on a logarithmic scale >1ps to clarify early time spectral dynamics.
16 Page 17 of 43 Physical Chemistry Chemical Physics
Table 1. Summary of fit parameters obtained by global analysis of TA measurements for 1c, 2c, and 3c in acetonitrile and cyclohexane.
Acetonitrile Cyclohexane
relax (ps) decay 1 (ps) decay 2 (ps) vib. (ps) relax (ps) decay 1 (ps) decay 2 (ps) vib. (ps) 1c 0.5 8.9 - - 0.48 4.6 25.7 - 2c 0.38 2.5 - 7.6 0.57 5.3 - 10.6 3c 0.33 3.1 - 18.0 - 1.4 6.2 13.9
In contrast, global fitting of data collected for 2c and 3c required a four-state series kinetic
model. Global fits yield state interconversion timescales of 0.38 ps, 2.5 ps, and 7.6 ps for 2c and
0.33 ps and 3.1 ps, and 18 ps for 3c. As for 1c, we attribute the shortest timescale with relaxation
within the excited state(s), while the intermediate timescale is associated with excited-state decay.
Notably, there is a significant decrease in excited-state lifetime between 1c and 2c-3c. As indicated
above, we assign the longest timescale obtained for 2c and 3c with decay of a vibrationally hot
state (흉풗풊풃. ) of the fraction of the switch population that relaxes back to the electronic ground state
without opening. Our assignment of this longer timescale to vibrational relaxation is supported by
the shape of SAS3: there is extra absorption to the red of the GSB (550 nm), where we expect
absorption of a vibrationally hot switch in its ground state to appear, but lacks significant
absorption appearing above 600 nm and below 500 nm that appear in SAS1 and SAS2 and is
attributable to excited-state absorption. It is interesting that a clear signature of a vibrationally hot
ground state is not captured for 1c, which likely reflects that vibrational cooling occurs on a
timescale that is comparable to or faster than electronic deactivation of 1c.
3.2.2 Solvent dependence of excited-state dynamics (1c-3c).
17 Physical Chemistry Chemical Physics Page 18 of 43
Prior ultrafast investigations of cycloreversion dynamics were conducted in various solvents, and we therefore used both acetonitrile and cyclohexane to facilitate comparisons with reports in the literature.47 Ultrafast transient spectra, as well as results from global fits for transient data collected with 1c-3c in cyclohexane are reported in the Supporting Information (Figure S10-
S15). Fit parameters are summarized in Table 1. 1c exhibits pronounced differences in spectral dynamics in cyclohexane vs. acetonitrile. Notably, there is an overall shift in the major absorption feature at ~698 towards ~710 nm. Global fitting of spectral dynamics for 1c in cyclohexane requires a four-component sequential model and reveals an additional lifetime associated with absorption decay. The global fit yielded a short lifetime of 0.48 ps (흉풓풆풍풂풙 ) associated with early time spectral evolution (i.e. shifts), as well as 4.6 (흉풅풆풄풂풚 ퟏ ) and 25.7 ps (흉풅풆풄풂풚 ퟐ ) lifetimes associated with evolution in the intensity of excited-state absorption features. These results are in close agreement with the results reported by Shim et al., who reported results for 1c in n-hexane including a major absorption feature at 710 nm and excited-state lifetimes of 4 and 22 ps.46
Like 1c, 2c and 3c both show a ~10 nm shift in spectral features towards lower energy from acetonitrile to cyclohexane. However, 2c does not exhibit an additional excited-state signal decay lifetime in cyclohexane while 3c does: 2c excited-state signal decay is slowed in cyclohexane from
2.5 to 5.3 ps, while the excited state signal decay for 3c in cyclohexane is slowed from a single 3.1 ps timescale to a bimodal decay with lifetimes of 1.4 and 6.2 ps. Vibrational cooling of ground- state 2c and 3c after excited-state deactivation is observable in both solvents. In summary, all three switches exhibit a dilation by a factor of ~2.1-2.8 of the excited-state lifetimes in cyclohexane vs. acetonitrile, suggesting that the impact of solvent environment on deactivation dynamics is comparable regardless of the core structure. Curiously, the bimodal decays for 1c and 3c in cyclohexane reflect changes in the timescales of nuclear dynamics on the excited-state potential
18 Page 19 of 43 Physical Chemistry Chemical Physics
energy surface in cyclohexane; these are discussed further below in context of previous
interpretations of ultrafast spectral dynamics collected with similar systems.
3.2.3 Temperature-dependent excited-state lifetimes (1c-3c).
Temperature-dependent TA measurements focused on the variation in the excited-state
lifetime of each switch dissolved in acetonitrile. These lifetimes are plotted in Figure 4; examples
of corrected transient data are presented in Figure S16. Noticeably, the excited-state lifetime of
1c is consistently much longer than those of 2c and 3c at all temperatures, while the excited-state
lifetime of 3c is only slightly longer than that of 2c. Further analysis of these data in combination
with temperature-dependent quantum yields are presented below.
Figure 4. Temperature-dependent excited-state lifetimes of 1c-3c in acetonitrile performed at temperatures in the range of 253-323 K.
3.2.4 Pendant-group dependence of excited-state dynamics in thienothiophene-core
switches (2c,4c,5c).
19 Physical Chemistry Chemical Physics Page 20 of 43
Figure 5. Transient absorption spectra collected for 4c (a), 5c (b) in acetonitrile at selected time delays ranging from 0.5ps to 100 ps (for 4c) and 1000 ps (for 5c) as well as the corresponding SAS as determined through global fitting for 4c (c), and 5c (d). SAS 3 and 4 in (c) have been scaled as indicated; (c) also includes an error bar (same scale as SAS 4) that represents the S/N level for our measurements). Data was normalized to 1 at maximum absorption at 1ps.
We also explored the dependence of excited-state dynamics in TT-core switches with variation in pendant group to enable further comparisons with perfluorocyclopentene-core switches. Figure 5a presents the transient spectra for 4c following excitation at 600 nm. 4c exhibits a broad excited-state absorption feature spanning the visible with overlapping GSB features at 400 nm and 650 nm. Spectra collected at long pump-probe delays (~10 ps) exhibit signatures of a vibrationally hot ground electronic state, as observed for 2c and 3c. The spectral data after 0.5 ps was fit to a four-component sequential global fitting model, resulting in timescales associated with
20 Page 21 of 43 Physical Chemistry Chemical Physics
excited-state signal decay 0.5 ps (흉풅풆풄풂풚ퟏ ), excited-state decay 2 ps (흉풅풆풄풂풚 ퟐ ), as well as a
vibrational relaxation on the ground state 12 ps (흉풗풊풃. ). For 4c the lifetime for SAS 1 needed to be
constrained to a minimum of 0.5 ps in order to minimize the residuals and improve the fit quality.
Figure 5c shows the SAS for 4c. SAS 3 and SAS 4 were scaled by factors of 5 and 40, respectively,
to highlight spectral shapes more clearly. As with 1c-3c, we expected to see a permanent bleach
signal in SAS 4 reflecting that some cycloreversion occurs. However, due to a very small
cycloreversion quantum yield for 4c (0.44%), the permanent bleach is barely distinguishable from
baseline noise, which we estimate is approximately 0.005 (normalized differential absorption,
relative to maximum absorption at 1 ps) based on baseline fluctuations for signals collected well
before time zero. Time dependent cuts taken at representative wavelengths are shown in Figure 6
(column a) at: 405 nm, 440 nm, 530 nm, and 645 nm. Global fits and residuals are presented in
Figure S17.
Figure 5b shows the transient spectra of 5c in acetonitrile following excitation at 530 nm.
5c exhibits a broad excited state absorption feature encompassing the visible spectrum with an
overlapping GSB at 540 nm. The excited state feature appears to decay rapidly leading to a
vibrational hot state in the ground electronic state as well as another absorption feature that we
attribute to a transient signal from a possible photoproduct. In contrast with 1-4, this photoswitch
was observed to be slightly unstable, forming a permanent photoproduct after prolonged exposure
to UV light. Although we were unable to isolate this photoproduct or its steady-state absorption
features, TAS signals are largely similar to those measured for other TT switches studied here,
indicating that contributing signals arising from the photoproduct are minor at best. As with 1c,
2c, and 3c, 5c also demonstrates a late time permanent bleach indicative of cycloreversion.
21 Physical Chemistry Chemical Physics Page 22 of 43
Transient spectra for 5c were fit with a four-component sequential global fitting model.
The SAS for 5c are shown below in Figure 5d. Time dependent global fits are shown in Figure 6
(column b) at representative wavelengths at 450 nm, 540 nm, 590 nm, and 680 nm, with global fits and residuals presented in Figure S18. The global fits revealed a short-lived electronic excited state (흉풅풆풄풂풚 ퟏ ) of 1.1 ps and a longer timescale of 8.6 ps associated with vibrational relaxation in the electronic ground state (흉풗풊풃. ). The global fit also revealed a second long-lived species (SAS3) with a 155 ps lifetime, which may be the result of exciting a photoproduct that also absorbs at 530 nm. The late time spectrum shows permanent bleaching from cycloreversion. The fit results are summarized in Table 2.
22 Page 23 of 43 Physical Chemistry Chemical Physics
Figure 6. Time-dependent global fits for 4c (column a) and 5c (column b) in acetonitrile. Data after 0.5 ps is plotted on a linear scale <1ps and on a logarithmic scale >1ps to clarify early time spectral dynamics.
23 Physical Chemistry Chemical Physics Page 24 of 43
Table 2. Summary of fit parameters for 4c and 5c in acetonitrile.
relax (ps) decay 1 (ps) decay 2 (ps) vib. (ps) photoproduct. (ps) 4c - 0.5 2 12 - 5c - 1.1 - 8.6 155
3.4 Cycloreversion Quantum yields.
Details of quantum yield determinations are presented in the Supporting Information (e.g.
Figures S21-S26). No significant differences in quantum yield for cycloreversion (co) were found
for 1-3 after irradiation with 496 nm light at room temperature. All demonstrated relatively high
values for co, with co of 1c (0.32) in close agreement with the values reported by Sumi at al.
14 (~0.297). 5c also exhibited a modest co of 0.12 following irradiation with 496 nm light. In
contrast, the co for 4c at 590 nm (peak absorption) was drastically lower with a value of only
0.0044. All co results are summarized below in Table 3. Temperature dependent quantum yields
were also measured for photoswitches 1-3 at temperatures varying from 253-323K. Results for the
temperature dependent quantum yield studies are reported in Figure 7.
Table 3. Summarized co results for 1-5c at room temperature.
co 1c 0.320.02 2c 0.340.03 3c 0.310.02 4c 0.00440.0003 5c 0.120.006
24 Page 25 of 43 Physical Chemistry Chemical Physics
Figure 7. co results for 1-3c at temperatures ranging from 253-323K in acetonitrile. Each was fit using a linear least squares regression to assist with interpolating the best fit quantum yields for each temperature.
4. Discussion:
The photophysical mechanism for cycloreversion of “normal-type” DAE photoswitches
(such as 1c-5c) can be summarized with the potential energy diagrams sketched in Figure 817, 22,
24, 39: Photoexcitation occurs to the optically accessible 11B state, with rapid crossing to the 21A
25 Physical Chemistry Chemical Physics Page 26 of 43
surface near the Franck-Condon region; the nascent 21A population subsequently relaxes towards a local minimum. A barrier associated with lengthening of the reactive C-C bond separates this local minimum (as well as a nonreactive CI with the 11A ground-state surface) from the conical intersection through which cycloreversion occurs with branching between both open and cyclized structures. Hence, excited-state deactivation nominally involves parallel nonadiabatic relaxation pathways. Given that a significant fraction of the excited-state population returns to the ground state of the cyclized structure through either pathway, spectroscopic signatures of vibrational cooling should be observed under circumstances of fast electronic deactivation. Below we discuss the behaviors of 1c-5c and analogous switches within the context of this model.
Figure 8. Schematic of potential-energy surfaces and mechanisms for the photoinduced relaxation of 1c-5c.
4.1 Dependence of deactivation dynamics with core group (in acetonitrile).
Based on results from our transient absorption experiments, the photophysical dynamics of photoswitches 1c-3c are highly consistent with the picture shown in Figure 8: Each exhibits rapid
(sub-picosecond timescales ranging from 0.33-0.5 ps) spectral evolution that likely arise from nuclear relaxation on the 21A surface after an ultrafast transit through the 11B-21A intersection
26 Page 27 of 43 Physical Chemistry Chemical Physics
near the FC region; this assignment is supported by the fact that only modest changes in SAS
features are observed on this timescale from global fits. Based on previous studies, the 11B-21A
nonadiabatic transition of photoexcited normal-type DAE photoswitches is thought to occur within
~100 fs,29, 30, 39 a period that would be masked by coherent artifacts in our time-resolved data.
Subsequent deactivation from the 21A surface occurs on timescales of picoseconds, with lifetimes
dependent on the nature of core bridge, pendant groups, temperature, and solvent environment.
The reported 21A lifetime for 1c in acetonitrile (at room temperature) is 9.8 ps, which is highly
consistent with our observed value of 8.9 ps.47 In comparison, the switches with rigid core
structures (2c and 3c) deactivate from the 21A state on timescales of 2.5 ps and 3.1 ps respectively.
Finally, vibrational cooling is observed for 2c-3c, both of which have rapid excited-state (21A)
deactivation rates.
Based on our findings with 1c-3c, structural sensitivity of excited-state dynamics for this
series appears to be affected principally by core rigidity rather than aromaticity, as apparent by the
similar behavior of 2c and 3c and their dissimilar behavior compared to 1c: specifically, the
excited-state lifetimes of the rigid structures are comparable and consistently much shorter that of
1c in both high and low polarity solvents (acetonitrile and cyclohexane; dynamics in cyclohexane
are discussed in further below). We posit that structural rigidity likely impacts the shapes of the
21A potential-energy surface and facilitates rapid dynamics towards conical intersections through
which cycloreversion and nonreactive deactivation (“internal conversion”) occurs. Notably,
although the rigidity of the core structure impacts the excited-state lifetimes, it does not appear to
significantly impact the cycloreversion quantum yield: as shown in Table 3, the cycloreversion
quantum yields of 1c-3c fall within error of each other at room temperature (~0.32). This implies
that any changes to the potential energy surface and consequently nuclear dynamics that result
27 Physical Chemistry Chemical Physics Page 28 of 43
from structural modification has a roughly balanced impact on nonreactive and reactive
deactivation for this set of switches with benzothiophene pendant groups.
In order to better understand how the core bridging group impacts the shape of the 21A
potential-energy surface, we conducted a series of temperature-dependent measurements of the
cycloreversion quantum yields and 21A lifetimes for 1c-3c, similar in nature to a previous study
by Ishibashi et al. for 1c in n-hexane.32 We used these data to determine the relative rates for
cycloreversion and nonreactive deactivation and energy barriers associated with these pathways
based on their dependence with temperature. Assuming parallel kinetic deactivation of the 21A state, the rates for internal conversion (kIC) and cycloreversion (kO) can be determined using the total 21A decay rate and quantum yield. Similar to Ishibashi et al., we employed two approaches for determining these rates: The first approach assumes that all excited-switches that overcome the barrier to the cycloreversion CI will open. The rates for each process are then determined as
흓푶(푻) 풌푶(푻) = (1) 흉ퟐ푨(푻)
ퟏ 풌푰푪(푻) = ― 풌푶(푻) (2) 흉ퟐ푨(푻)
In the second approach we assumed that only 55% (훘) of the excited-state population that
overcomes this barrier will reopen, thereby accounting for branching between the closed and open
isomers at the cycloreversion CI. The rates determined by this method are:
흓푶(푻) 풌푶(푻) = (3) ퟎ.ퟓퟓ ∗ 흉ퟐ푨(푻)
ퟏ 풌푰푪(푻) = ― 풌푶(푻) (4) 흉ퟐ푨(푻)
28 Page 29 of 43 Physical Chemistry Chemical Physics
We adopted this branching ratio (determined by Ishibashi et al. for 1c) for all three switches as we
do not anticipate significant differences in the shapes of the excited-state potential energy surface
32 for the open structure. We find that the rate of internal conversion kIC, is consistently higher than
the rate of cycloreversion kO for all three photoswitches as determined without branching. This
can be expected because internal conversion must account for most of the excited-state
deactivation with this model. In contrast, when branching is included, kIC is significantly decreased
as there are two pathways for deactivation back to the ground state of the cyclized structure.
We used the temperature-dependence of these rates to determine the activation energies,
enthalpies, and entropies for the two deactivation pathways according Equations 5 and 6:
푬풂 퐥퐧 (풌(푻)) = 퐥퐧 (푨) ― (5) 푹푻
⧧ ⧧ 풌풃푻 ―휟푯 휟푺 퐥퐧 (풌(푻)) ― 풍풏( ) = + (6) 풉 푹푻 푹
Temperature dependent rates fit to the Arrhenius equation (Equation 5) are plotted in Figure 9;
plots fit to Equation 6 are presented in the SI (Figure S27). Thermodynamic activation parameters
determined from these fits are summarized in Table 5.
29 Physical Chemistry Chemical Physics Page 30 of 43
Figure 9 Arrhenius plots for 1c-3c in acetonitrile without (a-c respectively) and with (d-f respectively) branching between closed and open structures at the cycloreversion CI.
30 Page 31 of 43 Physical Chemistry Chemical Physics
Table 5. Thermodynamic activation parameters for 1c-3c in acetonitrile.
Without Branching With Branching ⧧ ⧧ ⧧ ⧧ Species EA H S EA H S kcal/mol kcal/mol kcal/mol K kcal/mol kcal/mol kcal/mol K
1kIC 2.0 1.4 -0.004 1.4 0.88 -0.007 2 kIC 1.5 0.93 -0.004 1.1 0.58 -0.006 3 kIC 1.4 0.80 -0.005 1.0 0.41 -0.003
1 kO 2.9 2.3 -0.003 2.9 2.3 -0.0014 2 kO 2.0 1.4 -0.003 2.0 1.4 -0.0019 3 kO 2.1 1.5 -0.004 2.1 1.5 -0.003
Our results for 1c are consistent with those reported previously by Ishibashi et al. Those
authors report activation energies for 1c in n-hexane as 1.3 kcal/mol for IC and 4.2 kcal/mol for
cycloreversion when branching at the cycloreversion CI is included.32 While we find a similar
value for IC (1.4 kcal/mol), we obtain a barrier for cycloreversion that is more than 1 kcal/mol
lower in energy (2.9 kcal/mol). This difference could be the result of solvent stabilization effects
on the shape of the 21A surface, as the barrier to the cycloreversion CI is associated with an avoided
surface crossing; alternatively, this could be associated with differences in solvent-induced barriers
for structural reorganization. This difference in activation barrier is consistent with net slower
deactivation dynamics in cyclohexane vs. acetonitrile observed here and discussed in detail below.
Based on these analyses, we find that 1c exhibits higher activation energies and enthalpies
for both deactivation pathways than 2c and 3c when analyzed with and without branching at the
cycloreversion CI. It is most plausible that the inherent flexibility of the perfluorocyclopentene
core group allows for structural relaxation to a lower-energy local minimum on the excited-state
surface than is accessible to the rigid core photoswitches. In contrast, rigid core structures, like 2c
31 Physical Chemistry Chemical Physics Page 32 of 43
and 3c, have shallower potential-energy minima on the 21A surface due to tighter constraints on structural relaxation. Consequently, we would expect activation barriers to both internal conversion and cycloreversion that are comparatively lower for a switch with a rigid compared to flexible core bridge. Importantly, barriers to both internal conversion and cycloreversion are reduced with the rigid vs. flexible core structure, meaning that cycloreversion yields should remain fairly similar (as observed). In all cases the entropies of activation are negative, consistent with reaching a fairly specific excited-state structure (or set of structures) on the 21A surface (e.g. transition state or CI) prior to deactivation.
4.2 Dependence of deactivation dynamics with pendent group (TT core switches in acetonitrile).
In contrast, we find that the cycloreversion QY varies considerably with the nature of the pendent groups on the TT core. This variation is similar to what has been reported previously with the perfluorocyclopentene core and that has been ascribed to how the chemical nature of the pendent groups alters the avoided crossing between states of A symmetry that nominally results in the barrier that separates nonreactive from reactive geometries on the 21A surface:17-20 specifically, electron-withdrawing pendent groups destabilize the 21A surface of the cyclized isomer, lowering the barrier height and increasing the yields for cycloreversion; in contrast, electron-donating pendent groups stabilize the 21A state of the cyclized isomer, resulting in a larger barrier and smaller cycloreversion yield. For the set of structures studied here, we find no substantial differences in QY between photoswitches with the same pendent groups but different cores: In addition to the similarities for 1c-3c, Nakamura reports that the QY for the perfluorocyclopentene-core analog of 5c is 0.12, which matches the QY we observed for our thienothiophene core photoswitch.17
32 Page 33 of 43 Physical Chemistry Chemical Physics
We compare the time-resolved spectroscopy of 2c, 4c and 5c to obtain a global picture of
the cycloreversion dynamics’ dependence on pendent group. All exhibit fast (< 3 ps) excited-state
decay in acetonitrile. 2c undergoes fast relaxation in the 21A state on a sub-picosecond timescale
followed by excited state decay with a 2.5 ps lifetime. 4c differs slightly in that it appears to have
two excited-state signal decay timescales (sub-ps and 2.0 ps). The sub-ps decay also appears in the
recovery of ground-state bleach below 425 nm, implying bimodal excited-state deactivation,
although the origin of this behavior is not clear. We are unable to resolve sub-ps spectral dynamics
for 5c, suggesting that relaxation on the 21A surface occurs very rapidly and on a timescale that is
obscured by coherent artifacts in our measurement. The 21A excited state decays on a fast 1.1 ps
timescale which is consistent with what we observe for the other rigid core photoswitches.
The differences in QY and lifetimes for TT-core switches with different pendent groups
allows us to compare the relative rates for kO vs kIC at room temperature. For 2c, which has a
relatively high quantum yield, these rates are relatively similar: 0.056 ps-1 and 0.044 ps-1
-1 respectively. 4c on the other hand exhibits drastically different rates of 0.004 ps for kO and 0.50
-1 -1 -1 ps for kIC. 5c decays with rates of 0.22 ps for kO and 0.78 ps for kIC. While the QYs for
cycloreversion follow the prediction with electron donating/withdrawing character of the pendent
groups, it is interesting to note that the absolute rates for these processes vary considerably between
structures.
4.3 Dependence of deactivation dynamics with solvent environment.
Excited-state lifetimes of cyclized perfluorocyclopentene-core photoswitches in nonpolar
solvents (n-hexane or cyclohexane) have been reported to range between 9 and 25 ps depending
on the nature of pendent groups: Ward and Elles determined that the excited state of 1,2-bis(2,4-
33 Physical Chemistry Chemical Physics Page 34 of 43
dimethyl-5-phenyl-3-thienyl)perfluoro-cyclopentene decays on a 9 ps timescale.29, 30 Shim et al.
and Ishibashi et al. both determined the excited-state lifetimes for 1c and found that for non-polar solvents the excited state decays on a timescale of approximately 22-25 ps.46, 47 We similarly observe a 25 ps timescale in cyclohexane. Sotome et al. observed an excited-state lifetime of 12 ps for bis(2-methyl-5-phenylthiophen-3-yl)perfluorocyclopentene.39, 40 Ishibashi et al. reports a lifetime for 1c in acetonitrile as 9.8 ps47, which is a close match for our measurements in which we determined an excited state lifetime of 8.9 ps. We are not aware of any reported lifetimes for related switches in acetonitrile.
Spectral dynamics observed on intermediate timescales (3-4 ps) in nonpolar solution environments, as observed for 1c in cyclohexane, has been rationalized with more than one interpretation; how they are altered in polar solution environments is not well understood.
Relaxation timescales of photoexcited cyclized switches have been assigned to different mechanisms in the literature: Ward and Elles attributed a similar timescale observed in TAS and pump-repump-probe measurements with 1,2-bis(2,4-dimethyl-5-phenyl-3-thienyl)perfluoro- cyclopentene in cyclohexane to the barrier crossing on the way to the CI for cycloreversion; this
1 assignment is predicated on the theoretical expectation that a barrier exists between the 2 AC
1 1 1 29, 30 minimum and the 2 A/1 A CI (the 2 AO region in Figure 8). These authors noted that the ground state does not recover on the faster (3-ps) timescale for their switch29. If this intermediate timescale is associated with barrier crossing depicted in Figure 8, it would also most likely correspond to the timescale over which nonreactive deactivation occurs. This would imply that all nonreactive deactivation occurs via the region of the cycloreversion CI (i.e. at elongated C-C bond-length).
34 Page 35 of 43 Physical Chemistry Chemical Physics
In contrast, Sotome et al. and Shim et al. attributed this timescale to structural and/or
conformational changes, and possibly vibrational cooling, of the closed isomer in the 21A state;
they assign the longer timescale to excited-state decay through parallel internal conversion and
cycloreversion pathways, where the rate of the latter is limited by passage over the aforementioned
potential barrier.39, 40, 46 This assignment is most strongly supported by near-IR TA data that shows
a weak stimulated emission at ~1100 nm that persists for the excited-state lifetime and that must
originate from a location on the 21A surface removed from the cycloreversion CI as well as the
supposition that cycloreversion occurs through the same CI as cyclization, which generally occurs
on timescales ranging 100s of fs to a few ps.39 A recent time-resolved stimulated Raman study
reported by Mathies revealed signatures of structural dynamics on a timescale of a few picoseconds
(faster than excited-state deactivation) that could be consistent with either mechanism.56
The observed dilation of the 21A excited-state lifetime for TT-core switches in cyclohexane
vs. acetonitrile is qualitatively consistent with previous reports of solvent dependence for
perflurocyclopentene-core switches.47 Furthermore, our data for 1c-3c in cyclohexane seem
consistent with the picture that an intermediate (few ps or less) timescale is associated with
structural or conformation dynamics as well as vibrational cooling in the 21A state. We do not
observe spectral dynamics on such a timescale for the rigid-core 2c in cyclohexane, although a
second rapid lifetime (~1 ps) associated with excited state spectral dynamics is observed for 3c. It
is possible that the short, excited-state decay lifetime associated with 3c in cyclohexane is in fact
rapid structural rearrangement on the excited state, which results in dynamics away from the
Franck-Condon region for the excited-state absorption transition that results in loss of excited state
signal. Relaxation dynamics on the 21A surface involving large-scale structural changes to the core
35 Physical Chemistry Chemical Physics Page 36 of 43
should be inhibited by structural constraints in 2c and 3c, and the overall faster excited-state relaxation for both is qualitatively consistent with this picture.
The relationship between solvent polarity and cycloreversion dynamics remains poorly understood for this class of switch. Based on our comparisons of energy barriers determined with temperature dependent data for 1c in acetonitrile with those by Ishibashi in n-hexane, it is possible that a strongly polar solvent may stabilize regions of the potential energy surface along the cycloreversion reaction coordinate (but not the IC coordinate). However, the reversion QY is roughly the same in both solvents, implying that this modification does not significantly impact competition between reactive and nonreactive deactivation channels; thus, the modifications to these barriers must be offset by, e.g. solvent dependence of Arrhenius prefactors. Similarly, the lack of observable spectral dynamics associated with excited-state relaxation or vibrational cooling in acetonitrile could mean that the rate of these processes is limited by the absence/presence of energy-accepting solvent responses (i.e. solvation).
We also consider the possibility that some of the observed solvent dependence may be related to differences in solvent viscosity rather than solvent polarity. At our lower temperature limit (253 K), the excited-state lifetimes measured in acetonitrile approach similar lifetimes to what we observe in cyclohexane (Figure 4). Notably, the viscosity of acetonitrile at low temperatures (e.g. 0.83 mPa s at 233 K)57 approaches that of cyclohexane at room temperature
(0.91 mPa s at 298 K)58, suggesting that excited-state deactivation is more strongly controlled by solvent viscosity rather than solvent polarity. However, these similarities are likely coincidental, as excited-state lifetimes reported by Ishibashi for 1c in n-hexane 47 (a low viscosity solvent, 0.292
mPa s at 300 K)59 are similar to lifetimes we observe in cyclohexane at the same temperature, but
should be expected to be more similar to what is observed with room-temperature acetonitrile if
36 Page 37 of 43 Physical Chemistry Chemical Physics
viscosity dominated the solvent dependence of excited-state dynamics. In sum, while it seems
very likely that local solvent polarity may fundamentally alter the nonradiative deactivation
dynamics, more work is required to understand these different behaviors and their dependence on
solvent polarity and viscosity.
5. Conclusion.
The systematic investigation of structural, solvent, and temperature dependence on
ultrafast dynamics of DAE derivative photoswitches provides new insight into how core and
pendant structural motifs impact cycloreversion rates and quantum yields. Ultrafast transient
absorption spectroscopy revealed a distinct trend, with rigid core 2c and 3c switches demonstrating
faster switching than the analogous flexible perfluorocyclopentene core photoswitch, 1c.
Cycloreversion quantum yields appear unaffected by core group structure as we observed similar
cycloreversion quantum yields for 1c-3c. Our temperature dependent data provides insight into
this phenomenon; activation energy and enthalpy barriers are lower (i.e. the excited-state surface
is shallower) for rigid relative to flexible core switches for both internal conversion and
cycloreversion resulting in no net change in the branching ratio for each pathway. Extension of
this study to include 4c and 5c illustrates the previously noted dependence of cycloreversion
quantum yield on pendent group structure, but with increased excited-state deactivation rates due
to impacts of TT-core rigidity. Core- and pendent-dependent trends are observed in both
acetonitrile and cyclohexane, although excited-state relaxation and deactivation timescales are
longer for each switch in cyclohexane; in combination with results of previous solvent-dependent
studies of 1c, the latter suggests solvent-induced modification to the electronic surfaces.
Consistent with previous reports, we attribute spectral dynamics on a ~4 ps timescale observed for
37 Physical Chemistry Chemical Physics Page 38 of 43
1c in cyclohexane to structural rearrangement on the 21A surface; these dynamics are less prominent in rigid-core structures, supporting previous assignments that these arise from nuclear relaxation/reorganization in the 21A state.
Conflicts of Interest
There are no conflicts of interest to declare.
Acknowledgements
Photophysical studies and synthesis of 2-5 were supported by NSF-CHE-1607821 (A.E.B &
J.D.T). Photophysical studies were also supported in part by a JHU Harry and Cleio Greer
Fellowship (C. R. H.). We would also like to acknowledge the kind assistance of Joshua Wilhide and the staff at the Molecular Characterization and Analysis Complex at University of Maryland,
Baltimore County.
Supplementary Information. Additional transient absorption spectra, their spectral analyses, and methods for determining molar absorptivities and quantum yields are reported.
References
1. Irie, M.; Fukaminato, T.; Matsuda, K.; Kobatake, S., Photochromism of Diarylethene Molecules and Crystals: Memories, Switches, and Actuators. Chem. Rev. 2014, 114 (24), 12174- 12277. 2. Irie, M., Diarylethenes for Memories and Switches. Chem. Rev. 2000, 100 (5), 1685- 1716. 3. Chen, Y.; Zeng, D. X.; Fan, M. G., Synthesis and Photochromic Properties of Functional Diarylethene with a 2,5-Dihydrothiophene Bridging Unit. Org. Lett. 2003, 5 (9), 1435-1437.
38 Page 39 of 43 Physical Chemistry Chemical Physics
4. Chen, S.; Li, W.; Zhu, W.-H., Novel Ethene-Bridged Diarylethene Photochromic Systems: Self-Assembly, Photoswitches, and Molecular Logic Gates. In Photon-Working Switches, Yokoyama, Y. N., Keitaro, Ed. Springer: Japan, 2017; pp 37-68. 5. Staykov, A.; Yoshizawa, K., Photochemical Reversibility of Ring-Closing and Ring-Opening Reactions in Diarylperfluorocyclopentenes. J. Phys. Chem. C 2009, 113 (9), 3826-3834. 6. Staykov, A.; Areephong, J.; Browne, W. R.; Feringa, B. L.; Yoshizawa, K., Electrochemical and Photochemical Cyclization and Cycloreversion of Diarylethenes and Diarylethene-Capped Sexithiophene Wires. ACS Nano 2011, 5 (2), 1165-1178. 7. Morimoto, M.; Kobatake, S.; Irie, M., Multicolor Photochromism of Two- and Three- Component Diarylethene Crystals. J. Am. Chem. Soc. 2003, 125 (36), 11080-11087. 8. Fukaminato, T.; Hirose, T.; Doi, T.; Hazama, M.; Matsuda, K.; Irie, M., Molecular Design Strategy toward Diarylethenes That Photoswitch with Visible Light. J. Am. Chem. Soc. 2014, 136 (49), 17145-17154. 9. Bleger, D.; Hecht, S., Visible-Light-Activated Molecular Switches. Angew. Chem. Int. Ed. 2015, 54 (39), 11338-11349. 10. Herder, M.; Schmidt, B. M.; Grubert, L.; Patzel, M.; Schwarz, J.; Hecht, S., Improving the Fatigue Resistance of Diarylethene Switches. J. Am. Chem. Soc. 2015, 137 (7), 2738-2747. 11. Zhou, J.; Guo, X.; Katz, H. E.; Bragg, A. E., Molecular Switching via Multiplicity-Exclusive E/Z Photoisomerization Pathways. J. Am. Chem. Soc. 2015, 137 (33), 10841-10850. 12. Young, J. D.; Honick, C. R.; Zhou, J.; Pitts, C. R.; Ghorbani, F.; Peters, G. M.; Lectka, T.; Tovar, J. D.; Bragg, A. E., Energy- and Conformer-Dependent Excited-State Relaxation of an E/Z Photoswitchable Thienyl-Ethene. Phys. Chem. Chem. Phys. 2019, 21 (26), 14440-14452. 13. Aloïse, S.; Sliwa, M.; Pawlowska, Z.; Réhault, J.; Dubois, J.; Poizat, O.; Buntinx, G.; Perrier, A.; Maurel, F.; Yamaguchi, S.; Takeshita, M., Bridged Photochromic Diarylethenes Investigated by Ultrafast Absorption Spectroscopy: Evidence for Two Distinct Photocyclization Pathways. J. Am. Chem. Soc. 2010, 132 (21), 7379-7390. 14. Sumi, T.; Takagi, Y.; Yagi, A.; Morimoto, M.; Irie, M., Photoirradiation wavelength dependence of cycloreversion quantum yields of diarylethenes. Chem. Commun. 2014, 50 (30), 3928-3930. 15. Shizuka, T.; Seiya, K.; Tsuyoshi, K.; Masahiro, I., Extraordinarily High Thermal Stability of the Closed-ring Isomer of 1,2-Bis(5-methyl-2-phenylthiazol-4-yl)perfluorocyclopentene. Chem. Lett. 2003, 32 (10), 892-893. 16. Hanazawa, M.; Sumiya, R.; Horikawa, Y.; Irie, M., Thermally irreversible photochromic systems. Reversible photocyclization of 1,2-bis (2-methylbenzo[b]thiophen-3- yl)perfluorocyclocoalkene derivatives. J. Chem. Soc., Chem. Commun. 1992, (3), 206-207. 17. Nakamura, S.; Uchida, K.; Hatakeyama, M., Potential energy surfaces and quantum yields for photochromic diarylethene reactions. Molecules 2013, 18 (5), 5091-103. 18. Irie, M.; Sakemura, K.; Okinaka, M.; Uchida, K., Photochromism of Dithienylethenes with Electron-Donating Substituents. J. Org. Chem. 1995, 60 (25), 8305-8309. 19. Morimitsu, K.; Kobatake, S.; Irie, M., Control of Cycloreversion Quantum Yields of Diarylethenes by Introduction of Substituents at the Reactive Carbons. Mole. Cryst. Liq. Cryst. 2005, 431 (1), 451-454.
39 Physical Chemistry Chemical Physics Page 40 of 43
20. Tatsumi, Y.; Kitai, J.-i.; Uchida, W.; Ogata, K.; Nakamura, S.; Uchida, K., Photochromism of 1,2-Bis(2-thienyl)perfluorocyclopentene Derivatives: Substituent Effect on the Reactive Carbon Atoms. J. Phys. Chem. A 2012, 116 (45), 10973-10979. 21. Pariani, G.; Quintavalla, M.; Colella, L.; Oggioni, L.; Castagna, R.; Ortica, F.; Bertarelli, C.; Bianco, A., New Insight into the Fatigue Resistance of Photochromic 1,2-Diarylethenes. J. Phys. Chem. C 2017, 121 (42), 23592-23598. 22. Nakamura, S.; Kobayashi, T.; Takata, A.; Uchida, K.; Asano, Y.; Murakami, A.; Goldberg, A.; Guillaumont, D.; Yokojima, S.; Kobatake, S.; Irie, M., Quantum yields and potential energy surfaces: a theoretical study. J. Phys. Org. Chem. 2007, 20 (11), 821-829. 23. Rode, M. F.; Sobolewski, A. L., Effect of Chemical Substituents on the Energetical Landscape of a Molecular Photoswitch: An Ab Initio Study. J. Phys. Chem. A 2010, 114 (44), 11879-11889. 24. Asano, Y.; Murakami, A.; Kobayashi, T.; Goldberg, A.; Guillaumont, D.; Yabushita, S.; Irie, M.; Nakamura, S., Theoretical Study on the Photochromic Cycloreversion Reactions of Dithienylethenes; on the Role of the Conical Intersections. J. Am. Chem. Soc. 2004, 126 (38), 12112-12120. 25. Boggio-Pasqua, M.; Ravaglia, M.; Bearpark, M. J.; Garavelli, M.; Robb, M. A., Can Diarylethene Photochromism Be Explained by a Reaction Path Alone? A CASSCF Study with Model MMVB Dynamics. J. Phys. Chem. A 2003, 107 (50), 11139-11152. 26. Perrier, A.; Aloise, S.; Olivucci, M.; Jacquemin, D., Inverse versus Normal Dithienylethenes: Computational Investigation of the Photocyclization Reaction. J. Phys. Chem. Lett. 2013, 4 (13), 2190-2196. 27. Ishibashi, Y.; Fujiwara, M.; Umesato, T.; Saito, H.; Kobatake, S.; Irie, M.; Miyasaka, H., Cyclization Reaction Dynamics of a Photochromic Diarylethene Derivative as Revealed by Femtosecond to Microsecond Time-Resolved Spectroscopy. J. Phys. Chem. C 2011, 115 (10), 4265-4272. 28. Tamai, N.; Saika, T.; Shimidzu, T.; Irie, M., Femtosecond Dynamics of a Thiophene Oligomer with a Photoswitch by Transient Absorption Spectroscopy. J. Phys. Chem. 1996, 100 (12), 4689-4692. 29. Ward, C. L.; Elles, C. G., Controlling the Excited-State Reaction Dynamics of a Photochromic Molecular Switch with Sequential Two-Photon Excitation. J. Phys. Chem. Lett. 2012, 3 (20), 2995-3000. 30. Ward, C. L.; Elles, C. G., Cycloreversion Dynamics of a Photochromic Molecular Switch via One-Photon and Sequential Two-Photon Excitation. J. Phys. Chem. A 2014, 118 (43), 10011- 10019. 31. Jean-Ruel, H.; Cooney, R. R.; Gao, M.; Lu, C.; Kochman, M. A.; Morrison, C. A.; Miller, R. J. D., Femtosecond Dynamics of the Ring Closing Process of Diarylethene: A Case Study of Electrocyclic Reactions in Photochromic Single Crystals. J. Phys. Chem. A 2011, 115 (45), 13158- 13168. 32. Ishibashi, Y.; Umesato, T.; Kobatake, S.; Irie, M.; Miyasaka, H., Femtosecond Laser Photolysis Studies on Temperature Dependence of Cyclization and Cycloreversion Reactions of a Photochromic Diarylethene Derivative. J. Phys. Chem. C 2012, 116 (7), 4862-4869.
40 Page 41 of 43 Physical Chemistry Chemical Physics
33. Ern, J.; Bens, A.; Martin, H. D.; Mukamel, S.; Schmid, D.; Tretiak, S.; Tsiper, E.; Kryschi, C., Femtosecond reaction dynamics of a photochromic dithienylethene derivative. J. Lumin. 2000, 87-89, 742-744. 34. Tamai, N.; Miyasaka, H., Ultrafast Dynamics of Photochromic Systems. Chem. Rev. 2000, 100 (5), 1875-1890. 35. Arruda, B. C.; Sension, R. J., Ultrafast polyene dynamics: the ring opening of 1,3- cyclohexadiene derivatives. Phys. Chem. Chem. Phys. 2014, 16 (10), 4439-4455. 36. Shibata, K.; Muto, K.; Kobatake, S.; Irie, M., Photocyclization/Cycloreversion Quantum Yields of Diarylethenes in Single Crystals. J. Phys. Chem. A 2002, 106 (1), 209-214. 37. Fukaminato, T.; Umemoto, T.; Iwata, Y.; Yokojima, S.; Yoneyama, M.; Nakamura, S.; Irie, M., Photochromism of Diarylethene Single Molecules in Polymer Matrices. J. Am. Chem. Soc. 2007, 129 (18), 5932-5938. 38. Kobatake, S.; Yamada, T.; Uchida, K.; Kato, N.; Irie, M., Photochromism of 1,2-Bis(2,5- dimethyl-3-thienyl)perfluoro- cyclopentene in a Single Crystalline Phase. J. Am. Chem. Soc. 1999, 121 (11), 2380-2386. 39. Sotome, H.; Nagasaka, T.; Une, K.; Morikawa, S.; Katayama, T.; Kobatake, S.; Irie, M.; Miyasaka, H., Cycloreversion Reaction of a Diarylethene Derivative at Higher Excited States Attained by Two-Color, Two-Photon Femtosecond Pulsed Excitation. J. Am. Chem. Soc. 2017, 139 (47), 17159-17167. 40. Sotome, H.; Nagasaka, T.; Une, K.; Okui, C.; Ishibashi, Y.; Kamada, K.; Kobatake, S.; Irie, M.; Miyasaka, H., Efficient Cycloreversion Reaction of a Diarylethene Derivative in Higher Excited States Attained by Off-Resonant Simultaneous Two-Photon Absorption. J. Phys. Chem. Lett. 2017, 8 (14), 3272-3276. 41. Murakami, M.; Miyasaka, H.; Okada, T.; Kobatake, S.; Irie, M., Dynamics and mechanisms of the multiphoton gated photochromic reaction of diarylethene derivatives. J. Am. Chem. Soc. 2004, 126 (45), 14764-72. 42. Luchita, G.; Bondar, M. V.; Yao, S.; Mikhailov, I. A.; Yanez, C. O.; Przhonska, O. V.; Masunov, A. E.; Belfield, K. D., Efficient Photochromic Transformation of a New Fluorenyl Diarylethene: One- and Two-Photon Absorption Spectroscopy. ACS Appl. Mater. Interfaces 2011, 3 (9), 3559-3567. 43. Ern, J.; Bens, A. T.; Martin, H. D.; Mukamel, S.; Tretiak, S.; Tsyganenko, K.; Kuldova, K.; Trommsdorff, H. P.; Kryschi, C., Reaction Dynamics of a Photochromic Fluorescing Dithienylethene. J. Phys. Chem. A 2001, 105 (10), 1741-1749. 44. Ern, J.; Bens, A. T.; Martin, H. D.; Mukamel, S.; Schmid, D.; Tretiak, S.; Tsiper, E.; Kryschi, C., Reaction dynamics of photochromic dithienylethene derivatives. Chem. Phys. 1999, 246 (1), 115-125. 45. Tani, K.; Ishibashi, Y.; Miyasaka, H.; Kobatake, S.; Irie, M., Dynamics of Cyclization, Cycloreversion, and Multiphoton-Gated Reaction of a Photochromic Diarylethene Derivative in Crystalline Phase. J. Phys. Chem. C 2008, 112 (30), 11150-11157. 46. Shim, S.; Joo, T.; Bae, S. C.; Kim, K. S.; Kim, E., Ring Opening Dynamics of a Photochromic Diarylethene Derivative in Solution. J. Phys. Chem. A 2003, 107 (40), 8106-8110. 47. Ishibashi, Y.; Umesato, T.; Fujiwara, M.; Une, K.; Yoneda, Y.; Sotome, H.; Katayama, T.; Kobatake, S.; Asahi, T.; Irie, M.; Miyasaka, H., Solvent Polarity Dependence of
41 Physical Chemistry Chemical Physics Page 42 of 43
Photochromic Reactions of a Diarylethene Derivative As Revealed by Steady-State and Transient Spectroscopies. J. Phys. Chem. C 2016, 120 (2), 1170-1177. 48. Peters, G. M.; Tovar, J. D., Pendant Photochromic Conjugated Polymers Incorporating a Highly Functionalizable Thieno[3,4-b]thiophene Switching Motif. J. Am. Chem. Soc. 2019, 141 (7), 3146-3152. 49. Skov, A. B.; Ree, N.; Gertsen, A. S.; Chabera, P.; Uhlig, J.; Lissau, J. S.; Nucci, L.; Pullerits, T.; Mikkelsen, K. V.; Nielsen, M. B.; Sølling, T. I.; Hansen, T., Excited-State Topology Modifications of the Dihydroazulene Photoswitch Through Aromaticity. ChemPhotoChem 2019, 3 (8), 619-629. 50. Smith, M. C.; Snyder, J. A.; Streifel, B. C.; Bragg, A. E., Ultrafast Excited-State Dynamics of ortho-Terphenyl and 1,2-Diphenylcyclohexene: The Role of “Ethylenic Twisting” in the Nonadiabatic Photocyclization of Stilbene Analogs. J. Phys. Chem. Lett. 2013, 4 (11), 1895-1900. 51. Fleming, C.; Remón, P.; Li, S.; Simeth, N. A.; König, B.; Grøtli, M.; Andréasson, J., On the use of diarylmaleimide derivatives in biological contexts: An investigation of the photochromic properties in aqueous solution. Dyes Pigments 2017, 137, 410-420. 52. Yu, W.; Magnanelli, T. J.; Zhou, J.; Bragg, A. E., Structural Heterogeneity in the Localized Excited States of Poly(3-hexylthiophene). J. Phys. Chem. B 2016, 120 (22), 5093-5102. 53. Molloy, M. S.; Snyder, J. A.; Bragg, A. E., Structural and Solvent Control of Nonadiabatic Photochemical Bond Formation: Photocyclization of o-Terphenyl in Solution. J. Phys. Chem. A 2014, 118 (22), 3913-3925. 54. Stranius, K.; Börjesson, K., Determining the Photoisomerization Quantum Yield of Photoswitchable Molecules in Solution and in the Solid State. Sci. Rep. 2017, 7, 41145. 55. Roibu, A.; Fransen, S.; Leblebici, M. E.; Meir, G.; Van Gerven, T.; Kuhn, S., An accessible visible-light actinometer for the determination of photon flux and optical pathlength in flow photo microreactors. Sci. Rep. 2018, 8 (1), 5421. 56. Valley, D. T.; Hoffman, D. P.; Mathies, R. A., Reactive and unreactive pathways in a photochemical ring opening reaction from 2D femtosecond stimulated Raman. Phys. Chem. Chem. Phys. 2015, 17 (14), 9231-9240. 57. Pugachev, Y. F., Absonderung der reinigen Acetonitril und gamma-Butyrlakton nach Zonenschmelzemethode und Untersuchung der Temperaturabhängigkeit von Viskositätskoeffizienten. Izv. Vyssh. Uchebn. Zaved. Khim. Khim. Tekhnol. 1991, 34, 111-113. 58. Dean, J. A., Lange’s Handbook of Chemistry. 15th ed.; McGraw-Hill: New York, NY, 1998; p 1424. 59. Michailidou, E. K.; Assael, M. J.; Huber, M. L.; Perkins, R. A., Reference Correlation of the Viscosity of n-Hexane from the Triple Point to 600 K and up to 100 MPa. J. Phys. Chem. Ref. Data 2013, 42 (3), 033104.
42 Page 43 of 43 Physical Chemistry Chemical Physics
TOC Graphic:
43